deepgeek/CosyVoice
2
0
Fork 0
mirror of https://github.com/FunAudioLLM/CosyVoice.git synced 2026-02-05 18:09:24 +08:00
Code Issues Packages Projects Releases Wiki Activity

add docker compose for streaming tts

Browse Source
This commit is contained in:
root
2025-10-08 17:20:04 +08:00
parent a019a2504e
commit 7cbd490253
2 changed files with 23 additions and 4 deletions
Download Patch File Download Diff File Expand all files Collapse all files

20
runtime/triton_trtllm/docker-compose.dit.yml Normal file
View File

@@ -0,0 +1,20 @@
services:
tts:
image: soar97/triton-cosyvoice:25.06
shm_size: '1gb'
ports:
- "8000:8000"
- "8001:8001"
- "8002:8002"
environment:
- PYTHONIOENCODING=utf-8
- MODEL_ID=${MODEL_ID}
deploy:
resources:
reservations:
devices:
- driver: nvidia
device_ids: ['0']
capabilities: [gpu]
command: >
/bin/bash -c "pip install modelscope && cd /workspace && git clone https://github.com/yuekaizhang/Step-Audio2.git -b trt && git clone https://github.com/yuekaizhang/CosyVoice.git -b streaming && cd CosyVoice && git submodule update --init --recursive && cd runtime/triton_trtllm && bash run.sh 0 3"

7
runtime/triton_trtllm/run_stepaudio2_dit_token2wav.sh
View File

@@ -1,9 +1,8 @@
#!/bin/bash #!/bin/bash
# Copyright (c) 2025 NVIDIA (authors: Yuekai Zhang) # Copyright (c) 2025 NVIDIA (authors: Yuekai Zhang)
export CUDA_VISIBLE_DEVICES=0 export CUDA_VISIBLE_DEVICES=0
# cosyvoice_path=/workspace/CosyVoice cosyvoice_path=/workspace/CosyVoice
cosyvoice_path=/workspace_yuekai/tts/CosyVoice stepaudio2_path=/workspace/Step-Audio2
stepaudio2_path=/workspace_yuekai/tts/Step-Audio2
export PYTHONPATH=${stepaudio2_path}:$PYTHONPATH export PYTHONPATH=${stepaudio2_path}:$PYTHONPATH
export PYTHONPATH=${cosyvoice_path}:$PYTHONPATH export PYTHONPATH=${cosyvoice_path}:$PYTHONPATH
@@ -89,7 +88,7 @@ if [ $stage -le 2 ] && [ $stop_stage -ge 2 ]; then
LLM_TOKENIZER_DIR=$huggingface_model_local_dir LLM_TOKENIZER_DIR=$huggingface_model_local_dir
BLS_INSTANCE_NUM=$bls_instance_num BLS_INSTANCE_NUM=$bls_instance_num
TRITON_MAX_BATCH_SIZE=1 TRITON_MAX_BATCH_SIZE=1
DECOUPLED_MODE=True DECOUPLED_MODE=True # Only streaming TTS mode is supported using Nvidia Triton for now
STEP_AUDIO_MODEL_DIR=$step_audio_model_dir/token2wav STEP_AUDIO_MODEL_DIR=$step_audio_model_dir/token2wav
python3 scripts/fill_template.py -i ${model_repo}/${token2wav_dir}/config.pbtxt model_dir:${STEP_AUDIO_MODEL_DIR},triton_max_batch_size:${TRITON_MAX_BATCH_SIZE},max_queue_delay_microseconds:${MAX_QUEUE_DELAY_MICROSECONDS} python3 scripts/fill_template.py -i ${model_repo}/${token2wav_dir}/config.pbtxt model_dir:${STEP_AUDIO_MODEL_DIR},triton_max_batch_size:${TRITON_MAX_BATCH_SIZE},max_queue_delay_microseconds:${MAX_QUEUE_DELAY_MICROSECONDS}